// merge a Date entity and a Time entity, like "明天早上七点"
public List <Token> MergeDateAndTime(string text, DateObject referenceTime)
{
var ret = new List <Token>();
var ers = DatePointExtractor.Extract(text, referenceTime);
if (ers.Count == 0)
{
return(ret);
}
ers.AddRange(TimePointExtractor.Extract(text, referenceTime));
if (ers.Count < 2)
{
return(ret);
}
ers = ers.OrderBy(o => o.Start).ToList();
var i = 0;
while (i < ers.Count - 1)
{
var j = i + 1;
while (j < ers.Count && ers[i].IsOverlap(ers[j]))
{
j++;
}
if (j >= ers.Count)
{
break;
}
if (ers[i].Type.Equals(Constants.SYS_DATETIME_DATE) && ers[j].Type.Equals(Constants.SYS_DATETIME_TIME))
{
var middleBegin = ers[i].Start + ers[i].Length ?? 0;
var middleEnd = ers[j].Start ?? 0;
if (middleBegin > middleEnd)
{
break;
}
var middleStr = text.Substring(middleBegin, middleEnd - middleBegin).Trim().ToLower();
if (string.IsNullOrEmpty(middleStr) || middleStr.Equals(",") || PrepositionRegex.IsMatch(middleStr))
{
var begin = ers[i].Start ?? 0;
var end = (ers[j].Start ?? 0) + (ers[j].Length ?? 0);
ret.Add(new Token(begin, end));
}
i = j + 1;
continue;
}
i = j;
}
return(ret);
}
public List <Token> TimeEveryday(string text, DateObject referenceTime)
{
var ret = new List <Token>();
var ers = TimeExtractor.Extract(text, referenceTime);
foreach (var er in ers)
{
var beforeStr = text.Substring(0, er.Start ?? 0);
var match = EachDayRegex.Match(beforeStr);
if (match.Success)
{
ret.Add(new Token(match.Index, match.Index + match.Length + (er.Length ?? 0)));
}
}
return(ret);
}
public List <ExtractResult> Extract(string text, DateObject referenceTime)
{
var ret = DateExtractor.Extract(text, referenceTime);
// the order is important, since there is a problem in merging
AddTo(ret, TimeExtractor.Extract(text, referenceTime));
AddTo(ret, DurationExtractor.Extract(text, referenceTime));
AddTo(ret, DatePeriodExtractor.Extract(text, referenceTime));
AddTo(ret, DateTimeExtractor.Extract(text, referenceTime));
AddTo(ret, TimePeriodExtractor.Extract(text, referenceTime));
AddTo(ret, DateTimePeriodExtractor.Extract(text, referenceTime));
AddTo(ret, SetExtractor.Extract(text, referenceTime));
AddTo(ret, HolidayExtractor.Extract(text, referenceTime));
CheckBlackList(ref ret, text);
AddMod(ret, text);
ret = ret.OrderBy(p => p.Start).ToList();
return(ret);
}
private List <Token> MergeTwoTimePoints(string text, DateObject referenceTime)
{
var ret = new List <Token>();
var er1 = TimeWithDateExtractor.Extract(text, referenceTime);
var er2 = SingleTimeExtractor.Extract(text, referenceTime);
var timePoints = new List <ExtractResult>();
// handle the overlap problem
var j = 0;
for (var i = 0; i < er1.Count; i++)
{
timePoints.Add(er1[i]);
while (j < er2.Count && er2[j].Start + er2[j].Length <= er1[i].Start)
{
timePoints.Add(er2[j]);
j++;
}
while (j < er2.Count && er2[j].IsOverlap(er1[i]))
{
j++;
}
}
for (; j < er2.Count; j++)
{
timePoints.Add(er2[j]);
}
timePoints = timePoints.OrderBy(o => o.Start).ToList();
// merge "{TimePoint} to {TimePoint}", "between {TimePoint} and {TimePoint}"
var idx = 0;
while (idx < timePoints.Count - 1)
{
// if both ends are Time. then this is a TimePeriod, not a DateTimePeriod
if (timePoints[idx].Type.Equals(Constants.SYS_DATETIME_TIME) &&
timePoints[idx + 1].Type.Equals(Constants.SYS_DATETIME_TIME))
{
idx++;
continue;
}
var middleBegin = timePoints[idx].Start + timePoints[idx].Length ?? 0;
var middleEnd = timePoints[idx + 1].Start ?? 0;
var middleStr = text.Substring(middleBegin, middleEnd - middleBegin).Trim();
// handle "{TimePoint} to {TimePoint}"
if (TillRegex.IsExactMatch(middleStr, trim: true))
{
var periodBegin = timePoints[idx].Start ?? 0;
var periodEnd = (timePoints[idx + 1].Start ?? 0) + (timePoints[idx + 1].Length ?? 0);
// handle "from"
var beforeStr = text.Substring(0, periodBegin).ToLowerInvariant();
if (beforeStr.Trim().EndsWith("从"))
{
periodBegin = beforeStr.LastIndexOf("从", StringComparison.Ordinal);
}
ret.Add(new Token(periodBegin, periodEnd));
idx += 2;
continue;
}
// handle "between {TimePoint} and {TimePoint}"
if (middleStr.Equals("和") || middleStr.Equals("与") || middleStr.Equals("到"))
{
var periodBegin = timePoints[idx].Start ?? 0;
var periodEnd = (timePoints[idx + 1].Start ?? 0) + (timePoints[idx + 1].Length ?? 0);
// handle "between"
var afterStr = text.Substring(periodEnd).ToLowerInvariant();
var match = ZhijianRegex.Match(afterStr);
if (match.Success)
{
ret.Add(new Token(periodBegin, periodEnd + match.Length));
idx += 2;
continue;
}
}
idx++;
}
return(ret);
}